Portfolio project // Live demo
Made by Jeremy BurkeDetect. Understand.
Deploy with confidence.
Upload an image or pick a sample scene, run live object detection with seven models — YOLOv8 through YOLO12, RT-DETR, and open-vocabulary YOLO-World — through a Firebase + Cloud Run pipeline, and export the annotated image plus JSON and CSV outputs.
Next.jsTypeScriptTailwind CSSFirebaseCloud RunFastAPIUltralytics YOLORT-DETR
urban-intersection.jpgYOLO11n · 22 objects
motorcycle 71%
motorcycle 69%
car 66%
motorcycle 62%
25.0%
Ready
Sample scenes
Public urban-scene photos with precomputed YOLO11n detections
Loading precomputed detections…
Why I built this
This project demonstrates an end-to-end computer vision workflow, not just a static model output: a user-facing web app that accepts an upload, queues an inference job, processes it in the cloud, and returns usable detection artifacts — the way production AI tools are typically structured.
How the live pipeline works
- 1Upload an image from the dashboard
- 2The image is stored securely in Firebase Storage
- 3A Firestore job tracks inference status in real time
- 4Cloud Run processes the image with the selected YOLO/RT-DETR model
- 5Download the annotated image, JSON, or CSV results
What it demonstrates
- Next.js + TypeScript frontend development
- Firebase Auth, Storage, and Firestore integration
- Cloud Run FastAPI backend deployment
- Multi-model detection (YOLOv8–12, RT-DETR, YOLO-World)
- Asynchronous job status tracking
- Annotated image, JSON, and CSV generation